Article pubs.acs.org/JPCC
Compressibility of Nanocrystalline TiO2 Anatase Yahya Al-Khatatbeh,*,†,‡ Kanani K. M. Lee,‡ and Boris Kiefer§ †
Physics Department, Bard College at Simon’s Rock, Great Barrington, Massachusetts 01230, United States Department of Geology and Geophysics, Yale University, New Haven, Connecticut 06511, United States § Department of Physics, New Mexico State University, Las Cruces, New Mexico 88003-8001, United States ‡
ABSTRACT: Using high-resolution, synchrotron-based powder X-ray diffraction (XRD), we have studied the high-pressure behavior of the anatase phase of nanocrystalline TiO2 (nc-TiO2) under hydrostatic conditions. We find that for anatase with a grain size larger than ∼40 nm, the room-pressure bulk modulus K0 remains constant at ∼200 GPa to within experimental uncertainties. An ∼15% decrease in K0 is observed for grains that are ∼20 nm in size and remains unchanged for grains down to 6 nm in diameter, indicating a rapid increase in compressibility for nc-TiO2 anatase between 40 and 20 nm.
■
INTRODUCTION TiO2 anatase is an important material in many technological applications including photocatalysis, ceramics, electronic storage media, and hydrogen storage applications.1−3 Due to its promising technological applications,1−3 the high-pressure behavior of nc-TiO2 anatase has attracted great interest over the past decade4−17 with studies focusing on pressure-induced phase transitions and amorphization.5−7,9,11,12,14−17 However, little is known about the effect of particle size on the equation of state (EOS) of TiO2 anatase and the few available studies are inconclusive and controversial: previous experiments on ncTiO2 anatase have found that the bulk modulus may increase or decrease with decreasing particle size.4,8,10,13 Some of the complications likely arise from the lack of hydrostaticity in the sample under pressure,8,10 whereas other studies do not provide the EOS of nc-TiO2 anatase13 or have some important issues (e.g., typographical errors or mistakes in EOS determination) in the reported EOS and/or measured value of its zero-pressure volume V0.4 Thus, a consistent understanding of the behavior of nanocrystalline TiO2 anatase as grain size is changed is presently lacking. To eliminate, or at least reduce, the controversy regarding the grain size dependence, we study the size dependence of the bulk modulus in nc-TiO2 anatase under hydrostatic conditions at pressures up to ∼11 GPa. Therefore, this work along with a reanalysis of previous studies on nc-TiO2 anatase4,8,10,13 allow us to reach consistency among several studies of the size dependence of the bulk modulus of nc-TiO2 anatase.
diamond-anvil cell (DAC) experiments. We confirmed the grain sizes by using the full-width-half-maximum18 of the XRD peaks to be 40(±3) and 20(±4) nm, respectively, consistent with scanning electron microscope measurements which yielded particle sizes of ∼40−60 and ∼20−30 nm, respectively. We performed two independent room temperature, highpressure DAC experiments to study the EOS of nc-TiO2 anatase (Table 1). A mixture of methanol−ethanol−water (M-E-W) (16:3:1 by volume) was used as the pressure medium.19 For pressures up to ∼11 GPa, M-E-W is hydrostatic.20,21 Rhenium and stainless steel gaskets of initial thicknesses of 270 μm were precompressed to thicknesses of 30 μm. The sample, pressure calibrant ruby,22 and pressure medium were placed in a 120 or 150 μm sample chamber in the center of the gasket and compressed between a pair of matched 300 μm culet diamonds. We collected angular-dispersive XRD at room temperature (Figure 1) using a MAR CCD at the HPCAT-IDB beamline (λ = 0.39779 Å) at the Advanced Photon Source (APS) at Argonne National Laboratory, and a MAR345 imaging plate at the B2 beamline (λ = 0.49594 Å), Cornell High Energy Synchrotron Source (CHESS), Cornell University. The pressure−volume (P−V) data nc-TiO2 anatase phase was fit by using a second-order Birch−Murnaghan EOS (BM-EOS),23 where the first pressure derivative of the bulk modulus K0′ is fixed to 4 to more easily compare with previous studies (Table 1). Other details on the experimental methods can be found elsewhere.24−26
■
EXPERIMENTAL METHODS Nanocrystalline samples of 99% TiO2 anatase/rutile powder (Nanostructured & Amorphous Materials, Inc., 50 nm average particle size) and 99.7% TiO2 anatase powder (Alfa Aesar, 15 nm average particle size) were used as starting materials in our © 2012 American Chemical Society
Received: July 31, 2012 Revised: September 11, 2012 Published: September 11, 2012 21635
dx.doi.org/10.1021/jp3075699 | J. Phys. Chem. C 2012, 116, 21635−21639
The Journal of Physical Chemistry C
Article
Table 1. The EOSs for nc-TiO2 Anatase and the Associated Experimental Conditionsa K0′
pressure range (GPa)
pressure medium
ref
4
0−11
M-E-W (16:3:1)
this study
169(9)
4
0−11
M-E-W (16:3:1)
this study
34.04
243(3)
4
0−16
no medium
ref 10
Nonhydrostatic compression
6
NA
237(3)
4
0−18
fluorinert FC70-FC77 (1:1)
ref 8
Nonhydrostatic compression above∼2.5 GPa
6
34.18
176(15) 165(9) 165(8) 168(10) 171(12)
4 4 4 4 4
0−23 0−9 0−13 0−17 0−19
M-E-W (16:3:1)
ref 13
- The EOS is not given in this ref - We compute the EOS from the given P−V data - V0 is determined from the 5 d-spacings given in this ref
6.5 15
NA 33.81
199(1.5) 239(4.8) 245(4.2) 222(12)* 200(10)* 173(13)* 185(2.6) 211(5) 230(5.1) 237(1.5) 234(2.6) 228(6.3) 226(4.6) 215(5.7)
NA NA NA 4* 12.27* 4* NA NA NA NA NA NA NA NA
0−12 0−12 5−12
M-E (4:1)
ref 4
- V0 for the 15 nm grain size is too low and contradicts findings
5.3 (0.1) 4*
0−14
grain size (nm)
V0 (Å )
K0 (GPa)
40
34.06(0.04)
198(10)
20
34.14(0.05)
30−40
4 7.2 9 13.5 21.3 24 30.1 45 microcrystalline
3
34.14* NA NA NA NA NA NA NA NA 34.07
190(10) 201(21)*
notes
- K0′ is not given in this ref - The 5−12 GPa for the 15 nm grain size is where our P− V data agree with this ref and we have used our V0 for the 20 nm size to fit a second-order BM-EOS
NA NA NA NA NA NA NA NA M-E-W (16:3:1)
ref 27
a The EOSs were obtained from a second-order BM-EOS23 (with K0′ = 4). For comparison, we list other experimental results.4,8,10,13,27 Where the K0′ value is given as 4; this value is fixed for the calculation. 1σ uncertainties are given in parentheses. For values not given, NA is shown. An asterisk indicates a revised EOS, using a second-order BM-EOS.23
■
medium does not exceed 2.5 GPa,28,29 which is clearly visible as a kink in the corresponding P−V diagram at ∼2 GPa (Figure 2a). The nonhydrostatic conditions have an effect on the EOS because of the systematic error introduced in the determined pressure of the material.30−32 Additionally, in a comparison with available data on the compressional behavior of nc-TiO2 anatase for sizes as low as 6 nm4,13 (Figure 2a) and for other nc-TiO2 phases,33,34 no such kink is observed. Consequently, nonhydrostaticity is likely responsible for the significant overestimation of the bulk modulus.8,10 A recent study4 explored the effect of grain size on nc-TiO2 anatase on the bulk modulus under hydrostatic conditions and reported a critical size of 15 nm below which K0 decreases (Table 1, Figure 2b). However, the measured bulk moduli in this study4 require further analyses as K0 obtained from a BMEOS is very sensitive to both V0 and K0′.23 The reported value of V0 (Table 1, Figure 2a) is inconsistent with previous measurements of nc-TiO2 anatase,13,35,36 where V0 increases as grain size decreases; ref 4 finds the opposite behavior. Moreover, K0′ is not reported,4 thus a true comparison cannot be made on the size dependence of the bulk modulus (Table 1). However, the P−V data of the 15 nm grains (ref 4) agree with our 20 nm size at P > 4 GPa (Figure 2a). Therefore, if we
RESULTS AND DISCUSSION In all experimental runs, nc-TiO2 anatase was stable up to the maximum pressure in this study of ∼11 GPa (Figures 1−3). The sample with 20 nm average particle size shows an ∼15% reduction in bulk modulus as compared to the sample with an average particle size of 40 nm (Table 1, Figure 2). Our measured second-order BM-EOS23 for the 40 nm nc-TiO2 anatase gives a bulk modulus K0 = 198(10) GPa with V0 = 34.06(0.04) Å3/f.u.; whereas for the 20 nm size, we obtain a significantly lower value of K0 = 169(9) GPa with V0 = 34.14(0.05) Å3/f.u. (Table 1, Figure 2). In our study, the bulk moduli of nc-TiO2 anatase for both sizes are significantly smaller than those reported previously.8,10 The reported K0 values in these studies are 237(3) and 243(3) GPa when K0′ is fixed to 4 for 6 nm8 and 30−40 nm10 grains, respectively (Table 1). Thus, despite the significant grain size difference in these studies,8,10 the bulk modulus appears to be almost size independent and much higher than that of the microcrystalline value of ∼200 GPa27 (Table 1). However, both studies8,10 suffered from nonhydrostatic compression conditions (Figure 2). No pressure medium was used in ref 10, whereas in ref 8, a Fluorinert FC70-FC77 (1:1) mixture was used. In the latter study, the hydrostaticity limit of the pressure 21636
dx.doi.org/10.1021/jp3075699 | J. Phys. Chem. C 2012, 116, 21635−21639
The Journal of Physical Chemistry C
Article
Figure 1. XRD patterns at various pressures during compression of nc-TiO2 anatase of 20 nm grain size up to 10.9 GPa show the corresponding hkl values for anatase (AN) and rutile (RT).
fixed to 4 regardless of the choice of the pressure range used to fit the second-order BM-EOS23 (Table 1). In detail, K0 = 165(9) GPa for P-range up to ∼9 GPa, where the M-E-W pressure medium is still hydrostatic, and K0 = 176(15) GPa for the whole P-range up to ∼23 GPa (Table 1). Thus, despite the small kink observed at pressures of 10−12 GPa,13 likely due to freezing of M-E-W pressure medium,20,21 the bulk moduli are consistent with our value of 169(9) GPa (Table 1). Thus, we conclude that the bulk modulus of anatase does not change for grain sizes between 6 and 20 nm. It has been suggested that at this small grain size, nc-TiO2 anatase is too small to sustain high dislocation concentrations, which is consistent with the observed reduction of the bulk modulus.4 The surface of the nc-TiO2 particles consists most likely of undersaturated oxygen and/or titanium sites. Therefore, these sites may undergo reactions with the M-E-W pressure medium. However, we do not observe OH− contamination through asymmetry in the XRD peaks and this effect is likely not dominant. An alternative mechanism is that the thickness of the strained surface layer is sufficiently large that it can affect the bulk modulus directly.37 With decreasing particle size we observe an increased effective volume relative to the microcrystalline reference (Table 1) suggesting the presence of an extensive (average) strain. We attribute the unexpectedly large reduction in bulk modulus to a significantly strained outer shell of the particles. Assuming that the thickness of the strained shell depends weakly on the size of the particle, it follows as the volume fraction of the shell relative to the particle volume increases with decreasing particle size that the bulk modulus reduction should become more pronounced. However, if this trend continues for very small particle size, where the energetics of the strained shell may promote restructuring, remains unclear. The measured bulk modulus of microcrystalline bulk TiO2 anatase of ∼200 GPa27 (Table 1) is in good agreement with our K0 for the 40 nm grain size. Both, ours (40 nm) and previous bulk anatase,27 show similar compressional behavior (Figure 2b). Since the 40 nm anatase behaves like microcrystalline samples, we conclude that the bulk modulus of anatase is size-
use our V0 = 34.14 Å3 for the 20 nm size and fit a second-order BM-EOS23 for the high-pressure data of ref 4 (i.e., for P > 4 GPa), we obtain K0 = 173(13) GPa, in excellent agreement with our measured value (Table 1) and ∼28% lower than the reported values in ref 4 of K0 = 239−245 GPa for the 15 nm grain size. Hence, an increase of V0 by 1%, for consistency with our study and others,13,35,36 leads to a significant reduction of the bulk modulus. On the other hand, if we fit a second-order BM-EOS to the entire pressure range, we obtain a K0 = 222 GPa (Table 1), still an ∼8% reduction in comparison to the reported values. The reported K0 values are 215(5.7) and 234(2.6) GPa for 45 and 21.3 nm average particle size, respectively.4 For similar sizes, our bulk moduli are 198(10) and 169(9) GPa (Table 1). Notably, the K0 value for the 6.5 nm is ∼20% higher than the value we compute for the 6 nm size13 (Table 1). However, while the data given in ref 4 are insufficient for a detailed reanalysis of the EOSs for each particle size, we suspect that a significant reduction of K0, similar to that for the 15 nm particle size, may also occur. Thus, in comparison to ours and another recent study13 where hydrostatic conditions are present, K0′ is constrained and V0 is consistent with previous studies,35,36 ref 4 may significantly overestimate K0 for nc-TiO2 anatase, which is at least partly due to the choice of K0′ as well as V0 as discussed above. In a recent study on the hydrostatic compressional behavior of 6 nm TiO2 anatase,13 we find excellent agreement with our results for the 20 nm grain size (Figure 3, Table 1). In addition to the similar hydrostatic conditions, the larger V0 value reported in this study is consistent with ours and previous measurements of nc-TiO2 anatase35,36 (Table 1). The change in normalized lattice parameters (a/a0 and c/c0) with pressure in ref 13 is also in good agreement with ours (Figure 3b), where c is much more compressible than a. However, the focus of ref 13 was on the unusual compressional behavior of nc-TiO2 anatase but without an EOS determination. We, thus, have computed the EOS for the 6 nm anatase from ref 13 (Table 1) and find excellent agreement with our measured K0 for the 20 nm size; the computed value of K0 has a value of ∼170 GPa with K0′ 21637
dx.doi.org/10.1021/jp3075699 | J. Phys. Chem. C 2012, 116, 21635−21639
The Journal of Physical Chemistry C
Article
Figure 2. Pressure versus volume of one TiO2 unit for nanocrystalline anatase for 20 (squares) and 40 nm (triangles) sizes. The solid symbols indicate points under compression, and open symbols indicate points upon decompression. (a) For the 20 nm grain size, our high-pressure volumes are smaller than previous work shown8 (vertical bowties). V0 in ref 8 is not given, thus we assumed the same value as measured by ref 13 for the same grain size. V0 of 15 nm grain size4 (horizontal bowties) is smaller than our V0 and is inconsistent with V0 measurements of nc-TiO2 anatase.13,35,36 (b) Our highpressure volumes are smaller than previous work10 on 30−40 nm (inverted triangles). The compression behavior of our 40 nm grain size is similar to that of microcrystalline TiO2 (crossed circles27).
Figure 3. (a) The compression behavior of our 20 nm grain size is in good agreement with the 6 nm size previously observed13 (circles). (b) The change in the normalized unit cell parameters (a/a0 and c/c0) as a function of pressure. Refer to Figure 2 caption for symbol identification. Solid lines indicate the linear BM-EOS53 for a and c axes up to ∼11 GPa. The dashed curve is extrapolated from the linear BM-EOS fit53 and shows good agreement beyond the hydrostatic limit of M-E-W as well as with a previous study.13.
independent for grain sizes larger than ∼40 nm (Figure 4). On the other hand, for smaller grain sizes we find a much lower K0 as compared to the 40 nm sample, suggesting that in the size range of 20−40 nm, the compressibility of nc-TiO2 anatase rapidly changes by ∼15% (Figure 4). The observed decrease in K0 of TiO2 anatase as the particle size is reduced is consistent with previous observations for some other oxides and metals (e.g., ZrO2,38,39 SnO2,40 Al2O3,41 MgO,42 ZnO,43 and Ni44), although this is not a universal behavior (e.g., CuO45 and Fe46 are size-independent, whereas K0 increases as particle size is reduced in Fe2O3,47 Ag,48 and Au48). We should also note that although the bulk modulus of anatase decreases with grain sizes ∼40 nm (2) and grain sizes from 6 to 20 nm, the bulk modulus is size independent. Between region (1) and region (2), the bulk modulus decreases by ∼15%, indicating a strong dependence of bulk modulus on particle size in the size range between 20 and 40 nm.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Portions of this work were performed at HPCAT at the APS, Argonne National Laboratory, and at B2 at CHESS, Cornell University. HPCAT is supported by DOE-BES, DOE-NNSA, NSF, and the W.M. Keck Foundation. APS and CHESS are supported by DOE. We thank beamline scientists Jesse Smith, Zhongwu Wang, Yue Meng, and Stanislav Sinogeikin. We also thank George Amulele for experimental help. Portions of this work were funded by the Carnegie/DOE Alliance Center.
■
REFERENCES
(1) Chen, X.; Mao, S. S. Chem. Rev. 2007, 107, 2891−2959. (2) Griffin, K. A.; Pakhomov, A. B.; Wang, C. M.; Heald, S. M.; Krishnan, K. M. Phys. Rev. Lett. 2005, 94, 157204. (3) Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647− 1650. (4) Chen, B.; Zhang, H.; Dunphy-Guzman, A. K.; Spagnoli, D.; Kruger, M. B.; Muthu, D. V. S.; Kunz, M. Phys. Rev. B 2009, 79, 125406. (5) Flank, A.-M.; Lagarde, P.; Itie, J.-P.; Hearne, G. R. Phys. Rev. B 2008, 77, 224112. (6) Hearne, G. R.; Zhao, J.; Dawe, A. M.; Pischedda, V.; Maaza, M.; Nieuwoudt, M. K.; Kibasomba, P.; Nemraoui, O.; Comins, J. D. Phys. Rev. B 2004, 70, 134102. (7) Machon, D.; Daniel, M.; Pischedda, V.; Daniele, S.; Bouvier, P.; LeFloch, S. Phys. Rev. B 2010, 82, 140102. (8) Pischedda, V.; Hearne, G. R.; Dawe, A. M.; Lowther, J. E. Phys. Rev. Lett. 2006, 96, 035509. (9) Shul’ga, Y. M.; Matyushenko, D. V.; Golyshev, A. A.; Shakhrai, D. V.; Molodets, A. M.; Kabachkov, E. N.; Kurkin, E. N.; Domashnev, I. A. Tech. Phys. Lett. 2010, 36, 841−843. (10) Swamy, V.; Dubrovinsky, L. S.; Dubrovinskaia, N. A.; Simionovici, A. S.; Drakopoulos, M.; Dmitriev, V.; Weber, H.-P. Solid State Commun. 2003, 125, 111−115. (11) Swamy, V.; Kuznetsov, A.; Dubrovinsky, L. S.; Caruso, R. A.; Shchukin, D. G.; Muddle, B. C. Phys. Rev. B 2005, 71, 184302. (12) Swamy, V.; Kuznetsov, A.; Dubrovinsky, L. S.; McMillan, P. F.; Prakapenka, V. B.; Shen, G.; Muddle, B. C. Phys. Rev. Lett. 2006, 96, 135702. (13) Swamy, V.; Kuznetsov, A. Y.; Dubrovinsky, L. S.; Kurnosov, A.; Prakapenka, V. B. Phys. Rev. Lett. 2009, 103, 075505. (14) Swamy, V.; Muddle, B. C. J. Aust. Ceram. Soc. 2008, 44, 1−5. (15) Wang, Z.; Saxena, S. K. Solid State Commun. 2001, 118, 75−78. (16) Wang, Z.; Saxena, S. K.; Pischedda, V.; Liermann, H. P.; Zha, C. S. J. Phys.: Condens. Matter 2001, 13, 8317−8323. (17) Wang, Y.; Zhao, Y.; Zhang, J.; Xu, H.; Wang, L.; Luo, S.; Daemen, L. L. J. Phys.: Condens. Matter 2008, 20, 125224. (18) Guinier, A. X-Ray Diffraction In Crystals, Imperfect Crystals, and Amporphous Bodies; Dunod: Paris, France, 1956. (19) Fujishiro, I.; Piermarini, G. J.; Block, S.; Munro, R. G. In Proceedings of the 8th AIRAPT Conference; Backman, C. M., 21639
dx.doi.org/10.1021/jp3075699 | J. Phys. Chem. C 2012, 116, 21635−21639